Analysis of the electrical characteristics and surge protection of EHV transmission lines supported by tall towers

Electrical Power and Energy Systems 57 (2014) 358–365

Contents lists available at ScienceDirect

Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes

Analysis of the electrical characteristics and surge protection of EHV transmission lines supported by tall towers André Jinno G. Pinto a, Eduardo Coelho M. Costa b,⇑, Sérgio Kurokawa b, José Humberto A. Monteiro a, Jorge Luiz de Franco a, José Pissolato a a b

University of Campinas – Unicamp, High Voltage Laboratory, Campinas, SP, Brazil Universidade Estadual Paulista – Unesp, Faculdade de Engenharia de Ilha Solteira – Feis, Ilha Solteira, SP, Brazil

a r t i c l e

i n f o

Article history: Received 25 March 2013 Received in revised form 5 December 2013 Accepted 10 December 2013

Keywords: Tall transmission towers Frequency-domain analysis Electromagnetic transients Lightning performance Surge arresters

a b s t r a c t The electrical characteristics of a new 280-m-tall transmission line are analyzed. The geometrical distance of the phases, wires height and physical structure of the line are intrinsic associated with its electrical performance in steady state as well as in transient conditions. Thus, a frequency-domain analysis is proposed comparing the longitudinal and transversal electrical parameters calculated from the tall transmission line with the same parameters obtained for a conventional 440-kV line. An adapted method for the mutual parameters is proposed to calculate the shunt capacitances considering two dielectrics between the wires and ground (air and forest), taking into account that the new transmission system has not a corridor, as usually observed in conventional lines. A second analysis is performed based on both lines modeling and time-domain simulations using the EMTP. Electromagnetic transient simulations from an atmospheric impulse shows that the tall transmission line has a major transient overvoltage than a conventional line. In sequence, an alternative surge protection is proposed using metal-oxide arresters connected directly at the phases, in parallel with the line insulators. The main objective of this technical report is to evaluate the electrical characteristics of the new transmission system using tall structures and, by this means, to highlight the possible variations concerning its operation and transient response. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction A relatively new steel tower has been implemented for transmission lines and non-uniform sections along them. This technology was first developed for Chinese transmission lines and it is now being proposed for Brazilian transmission grid. Fig. 1 shows a sample of a double-circuit transmission tower developed in China [1]. This structure is similar to those used for the transmission line crossing the Yangtze River, in the Jiangsu province, with a tower height of 349 m, spaced 2.3 km from each other. Currently, in accordance with the manufacturer’s database, some steel towers in China are 370 m. These structures are equipped with a cylindrical elevator that runs to the top of the tower [1]. In 2008, a transmission line with similar structure was proposed between Tucuruí/Macapá and Manaus, through the Amazon territory, in Northern Brazil. The proposed link is approximately 1850 km long and supported by several dozens of towers, between ⇑ Corresponding author. E-mail addresses: [email protected] (A.J.G. Pinto), [email protected] (E.C.M. Costa), [email protected] (S. Kurokawa), [email protected]. unicamp.br (J.H.A. Monteiro), [email protected] (J.L. de Franco), [email protected]. unicamp.br (J. Pissolato). 0142-0615/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijepes.2013.12.015

280 and 324 m high (Fig. 2), emphasizing that the conventional structures are 20 up to 50 m high [2,3]. The objective of this project is to link the Amazon state, maintained basically by obsolete thermoelectric power plants fed by diesel, to the Brazilian interconnected system, benefitting approximately 2.2 millions of people [3]. However, the Amazon region is an important world heritage site, making the careful planning of all engineering projects in this area a crucial task. This new technology, proposed to minimize environmental impacts and to overcome the local natural barriers, represents one of the most complex engineering developments in Brazil. Several sections with tall steel towers have been constructed into the Amazon tropical forest and across floodplain areas, which represents a great challenge taking into account issues regarding the project implementation itself and logistic planning. Another important question on this non-conventional system is associated with the surge protection of the line and insulation coordination, which are further discussed in the current paper. Few researches have been published about this emerging power transmission technique. Thus, in this work, a thorough analysis of the electrical characteristics of this non-conventional line is presented as a function of the frequency and the soil resistivity. The electrical parameters calculated from the analyzed transmission system are compared with the

A.J.G. Pinto et al. / Electrical Power and Energy Systems 57 (2014) 358–365

359

by using metal-oxide surge arresters connected directly at the towers.

2. Transmission line electrical parameters Overhead transmission lines and their propagation characteristics are dependent of their electrical longitudinal and transversal parameters, expressed to multiconductor transmission systems by the impedance matrix [Z] and admittance matrix [Y]. The self and mutual impedance terms in [Z] are variable with the frequency, according with the line and tower geometric features (height, distance between phases, geometric mean radius of bundle conductors, etc.), the physical features of the conductors and soil characteristics (conductivity, magnetic permeability and electric permittivity). Most of these characteristics are also intrinsically associated with the self and mutual terms of [Y]. In the following sub-sections, a brief overview is presented describing the theoretical basis to calculate the self and mutual parameters of transmission lines and used to the frequency-domain analyses proposed in the paper. Fig. 1. Transmission lines supported by 350-m-tall towers.

2.1. Longitudinal impedances Firstly, the following system composed of two conductors, i and k, is introduced. In the current frequency-domain analysis, a conservative condition is assumed, the ground is considered perfectly conducting. Based on this statement, the well-know image method can be applied to calculate the external impedances and admittances with good accuracy into the frequency range studied, from 0.01 Hz up to 1 MHz (switching and most of the lighting return strokes). Thus, the image of the wires, represented by i0 and k0 , are fixed irrespective of the frequency of the propagation pulse, as described in Fig. 3 [8,9]. Emphasizing that the wire height from the ground hi is equal to hi’ (depth of the wire image i0 into the soil). The total self and mutual impedances of a multiconductor system can be expressed as follows:

Fig. 2. 280-m-tall Transmission tower in the Amazon forest.

same parameters obtained from a conventional 440-kV transmission line, supported by 36-m-height towers. Fig. 2 shows an aerial image of a 280-m-tall tower in the Northern Brazil with the phases and shield wires crossing over the Amazon forest. The methodology applied to the proposed frequency-domain evaluation is described and cited along the paper as well as its eventual approximations. The proposed methodology takes into account the skin effect in the wires, the earth return current and the self/mutual external impedances/admittances [4–7]. In addition, an adapted method is introduced to calculate the shunt and mutual capacitances considering two distinct dielectrics between the line and ground. The first electric permittivity is related to the air and the second permittivity represents the tropical forest under the transmission line [8]. In a second part of the paper, a time-domain approach is carried out based on electromagnetic transients simulations from the line section supported by the 280-m-tall towers and from a conventional 440-kV line. An accurate frequency-dependent method is proposed to model the phases, shield wires and towers. Based on this modeling, several propagation characteristics and possible overvoltages associated with the non-conventional line sections are properly evaluated and a possible surge protection is proposed

Z ii ðxÞ ¼ Zext ii ðxÞ þ Z skin ðxÞ þ DZ ii ðxÞ

ð1Þ

Z ik ðxÞ ¼ Zextik ðxÞ þ DZ ik ðxÞ

ð2Þ

The self and mutual impedances are represented by the terms Zextii and Zextik, respectively. The terms DZii and DZik are the self and mutual earth return impedances (soil effect). The term Zskin is the impedance due to the skin effect, conventionally denominated as internal impedance. The skin effect has a more accentuated influence at very low frequencies and usually is neglected to study high-voltage transmission systems. On the other hand, the soil effect is predominant at medium up to high frequencies [8].

Fig. 3. Geometric configuration of a multiconductor system.

360

A.J.G. Pinto et al. / Electrical Power and Energy Systems 57 (2014) 358–365

Therefore, based on the approach using the image method, the self and mutual external impedances can be obtained from the following expressions:



l0 2hi ln 2p ri

Zextii ¼ jx

Zextik ¼ jx

l0 2p

 ð3Þ

  Dik ln dik

Pii ¼ ð4Þ

The self and mutual external impedances are represented by a frequency-dependent inductive reactance, characterized by a constant inductance as a function of the line geometry, wires’ radium ri (or geometric mean radius, GMR) and the magnetic permeability l0. This approach is more than enough for the frequency range analyzed in this study. The self and mutual earth return impedances are calculated based on the Carson’s ground impedances expressions. Firstly, a formulation to calculate the influence of the ground return was developed by Carson and Pollaczek and these formulas were also used to evaluate transmission line parameters. Both provides similar results for the earth return parameters of overhead lines, but Pollaczek’s formula is more adequate for underground conductors or pipes [4,7]. This methodology was developed from the calculation of the axial electric field in the ground and then, based on the electric field, the magnetic field components were also obtained from the Maxwell’s curl equation. Thus, based on a few approximations, the self and mutual earth return impedances are given by the following improper integrals [7]:

DZ ii ðxÞ ¼

DZ ik ðxÞ ¼

jxl0

p jxl 0

p

Z

1 0

Z 0

1

calculate the shunt and mutual capacitances of the line, as a function of the line geometry and the electric permittivity e0. The first step to obtain the capacitance matrix is to calculate the potential matrix of the multiconductor line, which the self and mutual terms are expressed as follows:

2hi u

e pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi du u2 þ jxl0 rg þ u

ð5Þ

e2ðhi þhk Þu cos ðdik uÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi du u2 þ jxl0 rg þ u

ð6Þ

The term rg represents the soil conductivity and hk is the height of the wire k from the ground. Eqs. (5) and (6) are finally expanded in terms of infinite series, which represents the more practical solution for these integrals. Assuming a properly quantity of terms in the Carson’s series expansion, the self and mutual earth return impedances can be calculated with accuracy for a few megahertz [8]. The self parameter Zskin is the internal impedance of the wires. This impedance is related to the skin effect, which is resulting from the electromagnetic field inside of the conductors. Several authors have previously studied this phenomenon in solid conductors, given in terms of modified Bessel functions and as a function of the wire propagation function [5,6]. Usually, the skin effect is neglected to study transmission line parameters and electromagnetic performance of high-voltage power systems. Although, the current evaluation takes into account the skin effect in order to increase the reliability of the proposed study.

Pik ¼

1 2pe0 1 2pe0

ln

  2hi ri

ð7Þ

ln

  Dik dik

ð8Þ

Thus, from (7) and (8), the capacitance matrix can be calculated by the inverse potential matrix:

½C ¼ ½P1

ð9Þ

However, the 280-m-tall transmission line has not a corridor, which means that the line wires are above the tropical forest. This means that the gap between the wires and the ground is composed of two distinct dielectrics: the air e0 and the forest eF. Thus, the conventional formulation presented in (7)–(9) is not completely proper for the current study. Therefore, a procedure adapted from the image theory is developed based on the following representation [10]. From the descriptions given in Fig. 4, the self and mutual terms of the potential matrix are reformulated:

Pii ¼

Pik ¼

1 2pe0 1 2pe0

ln

    h  hF 2h 1 h þ hF þ ln 2peF h þ hF r i h  hF 

ln

dikF Dik DikF dik

 þ

1 2peF

 ln

DikF dikF

ð10Þ

 ð11Þ

Eqs. (7) and (8) are substituted by (10) and (11), respectively, to calculate the potential matrix for the non-conventional line. Thereafter, the capacitance matrix can be obtained from the usual matrix inversion given in (9).

Fig. 4. Modified image representation assuming two dielectrics.

2.2. Transversal admittances The self and mutual admittances of overhead transmission lines are represented by two electrical parameters: conductance and capacitance. However, in transmission line modeling, only the imaginary part of the complex transversal admittance is considered, which represents the self and mutual capacitances of the line [10]. Another approach, commonly assumed to calculate the transversal parameters of transmission lines, is that the ground conductivity is of infinite value when the ground is assumed to be perfect. Based on this condition, the image method can be also applied to

Fig. 5. Geometric representation of a conventional 440-kV transmission line.

A.J.G. Pinto et al. / Electrical Power and Energy Systems 57 (2014) 358–365

361

3. Frequency-domain analysis of the longitudinal and transversal electrical parameters In this section, the electrical parameters of a 280-m-tall transmission line and a conventional 36-m-tall line are calculated and further compared. The frequency range analyzed is from 0.01 Hz up to 1 MHz. The geometric characteristics of the conventional and of the non-conventional transmission systems are described in the following table and in Fig. 5. The bundled conductors of both line representations have a similar structure. They are composed of four Grosbeak conductors spaced 0.4 m from each other. The shield wires are EHWS-3/8’’ and implicitly included in the line modeling [11]. Another important technical description is the distance between two consecutive towers, which is 400 m for the conventional representation and 2 km for the line sections using the tall towers. More geometrical and technical features of the non-conventional line sections are given by few available literatures and datasheets [2,3].

Fig. 7. Earth return resistances of the conventional (curve 1) and non-conventional (curve 2) lines and internal resistance (curve 3).

3.1. Variation on the longitudinal parameters The evaluation of the longitudinal parameters is carried out taking into account a variable soil resistivity for three distinct values: 100, 1000 and 10,000 X m. The self and mutual resistances and inductances are calculated for both lines, the 280-m-tall line is referred as the non-conventional line and the 36-m-tall line is denominated as conventional. The self and mutual resistances of the conventional line are denoted by Rc and the non-conventional line resistance is given by Rn. These parameters are extracted from the self and the mutual impedance Eqs. (1) and (2) , respectively. The differences between the parameters calculated from the two lines are evaluated based on the relation Rc/Rn, as follows: A major variation is observed for low soil resistivities, as shown in curves 1a and 1b. At 1 MHz, the resistance of the conventional line is almost 8% greater than the non-conventional line. However, up to 10 kHz, these variations are less than 1%. These behaviors can be explained analyzing the partial resistances due to the soil and skin effects. As described in the literature, the internal inductance due to the skin effect is usually neglected in transmission line studies [8]. Such assertion can be proved in Fig. 7, where the resistance due to the skin effect is significant just for very low frequencies. Thus, the variations measured in Fig. 6 (for self and mutual resistances) are solely related to the soil effect, assuming that the external impedances Zextii and Zextik are composed of an imaginary part representing an inductive reactance and the internal resistances are similar for both lines (and neglected for frequencies above 10 Hz).

Fig. 6. Relationship Rc/Rn of the self (a) and the mutual (b) resistances for the soil resistivities: 100 X m (1); 1000 X m (2) and 10 kX m (3).

Fig. 8. Relationship Lc/Ln of the mutual inductances for the soil resistivities: 100 X m (1); 1000 X m (2) and 10 kX m (3).

Fig. 9. Mutual earth return inductances of the conventional (curve 1) and nonconventional (curve 2) lines. Mutual external inductances of the conventional (curve 3) and non-conventional (curve 4) lines.

An equivalent procedure is carried out to measure the variations of the self and the mutual inductances. The inductances associated with the conventional line are denoted as Lc while those related to the non-conventional line representation are indicated

Fig. 10. Relationship Lc/Ln of the self inductances for the soil resistivities: 100 X m (1); 1000 X m (2) and 10 kX m (3).

362

A.J.G. Pinto et al. / Electrical Power and Energy Systems 57 (2014) 358–365

by Ln. Thus, Fig. 8 shows the variations given by the relation Lc/Ln of the mutual inductances. The mutual inductances have a major variation for low soil resistivities at high frequencies, as described in Fig. 8. The mutual inductance of the conventional line is 6% up to 18% major than the same parameter of the non-conventional line representation, depending of the soil resistivity and frequency range. Differently of the self and mutual resistivities, the mutual inductance variations between the lines are due to the earth return inductance and external inductance associated with the impedances DZik and Zextik in (2), respectively. The mutual earth return inductances and the mutual external inductances of the two line sections are described in details as follows. Fig. 9 shows that the mutual external inductances are constant, as prior observed in (4), and the value related to the non-conventional line is slightly higher than the same parameter for the conventional line section. However, the mutual earth return inductances are predominant on the mutual external impedances throughout the frequency range. Fig. 9 showed that the earth return inductance profile of the conventional line is major than the same inductance calculated from the non-conventional line, which results a greater total mutual inductance for the conventional line than the non-conventional line representation throughout the frequency range analyzed, specially for high frequencies close of 1 MHz. In contrast with the total mutual inductance, the self impedances are practically similar for the two line sections. The variations measured as a function of the frequency and soil resistivity are explicit in Fig. 10. The results in Fig. 10 show that the total self inductance function, calculated from the conventional line section, is minor than the nonconventional representation less than 0.7% at 1 MHz, taking into account a low soil resistivity. Otherwise, the variations are practically zero. This behavior can be investigated from the following figure, describing the earth return inductance and the external inductance in parts, which compose the total self inductance (see Fig. 11). In contrast with the results calculated for the mutual inductance, the external components are significant in the total self inductances expressed in (1). Such as for the total mutual inductance calculation, the sum of the earth return inductance to the external inductance results the total self inductance. Thus, the sum of the curves 1 and 3 represents the total self inductance of the conventional line section and analogously for the non-conventional line, the sum of the curves 2 and 4 is the total self inductance. The inductance resulted from the skin-effect impedance is very small and usually is neglected in the total self inductance [8].

3.2. Variation on the transversal parameters Based on (10) and (11), the capacitance matrix can be calculated for both line sections. Considering eF = e0, the capacitance matrix of

Fig. 11. Self earth return inductances of the conventional (curve 1) and nonconventional (curve 2) lines. Self external inductances of the conventional (curve 3) and non-conventional (curve 4) lines.

the conventional line can be also obtained applying the classic methodology, using (7)–(9). Otherwise, assuming the relative permittivity from 2 up to 5, which are values close to the wood permittivity, then the procedure proposed in 2.2 can be considered to calculate the transversal parameters of the non-conventional line [10]. The capacitance matrix of the non-conventional line is calculated based on three distinct values of eF: 8.85 gF/km (e0), 17.71 gF/km and 44.27 gF/km [10]. Emphasizing that the first electric permittivity is neglecting the influence of the forest under the line, assuming just the air permittivity. Thus, the capacitance matrices of the non-conventional line, calculated based on these three electrical permittivities, are respectively given as follows:

2

10:22

6 ½C 1  ¼ 4 3:50

3:50 3:50 9:32

3

7 1:97 5 gF=km

ð12Þ

9:32

3:50 1:97 2

3 10:30 3:41 3:41 6 7 ½C 2  ¼ 4 3:41 9:44 1:87 5 gF=km 2

10:35

6 ½C 3  ¼ 4 3:35

3:35 3:35 9:51

ð13Þ

9:44

3:41 1:87

3

7 1:80 5gF=km

ð14Þ

9:51

3:35 1:80

The matrices C1, C1 and C3 are calculated with eF equal 8.85, 17.71, and 44.27 gF/km; respectively. Emphasizing that these values can vary significantly depending of the vegetation and weather conditions. The capacitance matrix calculated for the conventional line is calculated and expresses as follows:

2

11:29

6 ½C c  ¼ 4 2:83

2:83 2:83 10:95

2:83 1:15

3

7 1:15 5 gF=km

ð15Þ

10:95

From a general overview in (12)–(15), the variations in the capacitances are around 10%, between the values calculated from the nonconventional and from conventional lines. Considering three distinct values for eF, as indicated in (12)–(14), the variations in the non-conventional line capacitances are discrete, taking into account the electrical permittivities considered in this study. 4. Time-domain analysis A time-domain analysis shows to be an appropriate approach to evaluate the variations prior observed in the frequency domain. Thus, a study of possible overvoltages was performed taking into account a fast transient represented by an atmospheric impulse. This fast and impulsive signal covers a wide frequency range, including the band analyzed in first part of this paper. The atmospheric impulse is modeled as a double exponential current wave, characterized by a front time of 2.6 ls and a tail time of 65 ls. Usually, in agreement with the International Electrotechnical Commission (IEC), a normalized atmospheric impulse is represented by a 8/20 ls current wave [14]. However, a front-wave rise of 20 kA/ls (di/dt) given by a 2.6/65 ls current wave has been widely used to simulate atmospheric surges on transmission towers and also covers the entire frequency range evaluated in the previous frequency-domain analysis. Actually, this wave shape represents a more conservative situation because of the steeper front time, major front-wave rise and a longer tail time than the IEC-normalized atmospheric impulse [12]. The conventional and non-conventional transmission line sections are modeled using the J. Marti frequency-dependent model,

A.J.G. Pinto et al. / Electrical Power and Energy Systems 57 (2014) 358–365

363

available in the Alternative Transient Program (ATP) [4,13]. The transmission towers are also modeled taking into account their characteristic impedances and geometrical features described in the Table 1. Further descriptions of the time-domain modeling are given in the following subsections. The following time-domain results shows more accentuated overvoltage profiles on the non-conventional line section (as expected, based on the electrical parameters evaluation and electrical and structural characteristics of the tall line sections). The main causes and conclusions about this behavior will be further discussed. Based on these features, an alternative solution using ZnO surge arresters connected at the tall towers is proposed, simulated and evaluated. The surge arrester time-domain modeling is further described and modeled from a commercial arrester usually applied for the same proposal in conventional transmission systems, with the same nominal voltage and electrical characteristics of the non-conventional system. Fig. 12. Transmission tower modeling using transmission line segments.

4.1. Transmission line modeling The shield wires and phases of the lines are modeled using a frequency-dependent digital line model directly in the phase domain, the Marti’s line model available in the ATP. This model represents properly the frequency-dependent nature of untransposed overhead transmission lines by means of recursive convolutions over a wide frequency range [13]. Furthermore, the Marti’s model was designed and implemented in the most of the Electromagnetic Transient Programs (EMTP), such as the free version of the ATP. In computational models to study lightning performance of transmission lines, the shield wires should be included as well as towers and eventual nonlinear corona effects. Accordingly, the necessary computational model may differ in comparison with models applied for switching overvoltages and other operational transients of the system [14–16]. The proposed phases and shield wires are modeled as frequency-dependent line sections using the mentioned computational line model. On the other hand, the metallic structure of the towers is modeled by horizontal and transversal transmission line sections, as described in Ref. [16]. The horizontal sections of the towers are modeled as horizontal transmission line where the parameters are obtained from standard line formulas. The vertical sections are also calculated using a specific formulation derived from the standard line formulas. The mutual coupling between any two parallel vertical transmission line sections is also taken into account [16]. The transmission tower given in Fig. 5 is represented using transmission line segments as follows in Fig. 12. Based on the proposed tower modeling, the electromagnetic transient of each section can be expressed by the well-known Transmission Line Telegrapher’s Equations [16]:

The terms L(x), R(x) and C(x) are the per unit-length inductance, resistance and capacitance, respectively. The Eq(17) do not take into account the frequency variations of the line parameters. However, since the skin depth is very small at high frequencies, the approach presented in [16] provides accurate results for fast and impulsive transients. Thus, the formulation presented in the Section 2 can be also applied to calculate the transmission line sections of the tower. These formulas were originally obtained for infinitely long lines which mean that end effects are neglected. For the case of the shortest truss segments this may not hold. Nevertheless, for the travel and the rise times of the waveforms involved in lightning, tower arms behave as long wires antennas. In this case, reflection analysis can be performed considering only the transmission line behavior [16]. In Ref. [17], an interesting analysis is presented comparing the modeling of tall towers using transmission line and antenna theories. The same reference assumes that the principal propagation mode is transverse electromagnetic (TEM) and the voltage difference can be obtained between any pair of points in the same transversal plane. Although the TEM representation is an approximation, it provides a practical method to apply well-known transmission line theory to model transmission towers. Furthermore, the frequency/time-domain analyses prior presented in Ref. [17], in which tower modeling using line and antenna theories are thoroughly discussed, proves that the transmission line representation is accurate enough to model towers to study impulses with time to maximum up to 1 ls (typical of subsequent return strokes). Thus, assuming the atmospheric impulse with a front-wave time of 2.6 ls, the tall tower modeling using transmission line theory shows to be physically consistent.



@ v ðx; tÞ @iðx; tÞ ¼ RðxÞiðx; tÞ þ LðxÞ @x @t

ð16Þ

4.2. Metal-oxide surge arrester modeling



@iðx; tÞ @ v ðx; tÞ ¼ CðxÞ @x @t

ð17Þ

The first step to model the surge arrester for the line configuration is to calculate its nominal voltage. This technical characteristic

Table 1 Geometric characteristics of the line sections.

y1 y3 x23 yPR xPR

Conventional (m)

Non-conventional (m)

27.64 24.04 18.54 36.0 15.02

280 255 24.0 300 12.0

Fig. 13. IEEE model for metal-oxide surge arresters.

364

A.J.G. Pinto et al. / Electrical Power and Energy Systems 57 (2014) 358–365

is calculated as a function of the maximum system voltage and from the temporary overvoltage (TOV) capability at 60 Hz, for 1 and 10 s. For both systems compared in this paper, the maximum operation voltage is 462 kV and the TOV capabilities for 1 and 10 s are 1.4 p.u. and 1.25 p.u., respectively [18]. Based on the technical features presented in the last paragraph and according with the information available in a given manufacturer database, 360-kV ZnO surge arresters are usually suitable for a nominal 440-kV system with the maximum voltage operation and TOV calculated before. The tower surge arresters used for the conventional and the non-conventional lines are the same (since the nominal voltage of both systems is the same) and modeled from a real and commercial device made by a well-established manufacturer. This device is properly modeled using the IEEE reference surge arrester model, described in Fig. 13. The value of the elements L0, R0, L1, R1 and C are calculated based on the number of the varistor columns and the physical height of the surge arrester. The magnitude of the residual voltage from an atmospheric impulse (20 kA 8/20 ls current wave) and from a switching operation (2 kA 30/80 ls current wave) are necessary to calculate the nonlinear resistances A0 and A1. The residual voltages of the 360-kV surge arrester used in the current analysis are 856 kV for atmospheric impulse and 712 kV for switching operation. Based on the technical database provided by the manufacturer, all electrical-circuit elements of the proposed equivalent model can be obtained. More details describing the surge arrester modeling are step-by-step described in Ref. [19]. 4.3. Electromagnetic transient simulations A conventional and a non-conventional infinite line segment are considered according with the previous descriptions of the towers, distance between them and wires. The surge impedance of the towers and the current surge representing an atmospheric impulse are also previously described. Based on these technical descriptions the time-domain simulations are carried out from an atmospheric impulse applied on a steel tower located at the middle of the line segments. A critical and more conservative situation is simulated, considering a total failure in the line shield and insulation (backflashover). The transient overvoltages on the phases at the tower struck by the atmospheric impulse followed by backflashover are measured considering both line segments, with and without the presence of ZnO surge arrester. The voltage profiles described in Fig. 14 shows a proper performance of the 360-kV surge arrester connected directly at the towers of a conventional 440-kV transmission line. The voltage surge is maintained into the maximum voltage range supported by the system. On the other hand, Fig. 15 shows the performance of the nonconventional system. The overvoltage peak on the system is more than 1500 kV, two time greater than the voltage peak observed in the conventional system. The curve 2 shows that the 360-kV ZnO arrester is not able to maintain the system voltage below of the maximum operational

Fig. 14. Atmospheric surge at a phase of the conventional line section without arrester (1) and with arrester (2).

Fig. 15. Atmospheric surge at a phase of the non-conventional line section without arrester (1) and with arrester (2).

voltage of 462 kV, taking into account that both transmission lines are operating with nominal 440 kV. Although the line segments were modeled assuming an infinite length (to avoid wave reflections), successive wave reflections are observed between 10 and 30 ls in the voltage profile without the surge arresters at the towers (curve 1). The greater voltage profile observed in the transmission line section with the tall towers can be firstly attributed to the variations observed at high frequencies in the resistance parameters and mutual inductances of the line. However, the high voltage levels in the non-conventional line are also because of the major distance between towers, since the surge impedances of both towers are very close and the tower footing impedance was the same for both line models. 5. Conclusions Based on the analyses carried out in frequency domain, it is verified that the mutual inductances of the tall line section have a significant variation at 60 Hz, around 7% up to 9% lower than a conventional 440-kV line (depending of the soil resistivity). However, for frequencies up to 1 MHz, the same variation is from 6% up to 18%, depending of the frequency and soil resistivity. Another significant variation between the non-conventional and the conventional representations was observed in the self and mutual resistances, up to 8% (Rc/Rn) at 1 MHz and 100 O m. These differences in the electrical parameters are intrinsic associated with the height and geometry of the towers as well as physical characteristics of the system. The variations prior evaluated in the frequency domain were also partially reflected in the time-domain results by means of electromagnetic transient simulations from an atmospheric impulse. Major voltage peaks and reflections were observed in the simulation performed from the non-conventional line section. This behavior is mainly attributed to the minor self resistance of the wires of the non-conventional section, minor electromagnetic coupling among phases and shield wires (mutual inductance among phases and shield wires) and also because of the distance between two consecutive towers, which is five times major for the non-conventional line section compared to the conventional section. This last feature results less attenuation of the voltage surge along the line. Furthermore, the non-conventional line section is characterized by a minor number of towers per length, which results a few shunt connections to drain the electrical current surge to the ground. All these features are directly associated with the major voltage levels observed in the system composed of tall towers. The overvoltage levels observed in the line sections supported by tall tower are a critical issue because of two main factors. First, the towers are more than 200-m-heigth (considering the Brazilian and the Chinese towers), which means that these steel structures are tall enough to be close to storm clouds (very common in the Amazon Region and in the Northern Brazil). This way, the shield wires cannot provide an effective protection at the line flanks,

A.J.G. Pinto et al. / Electrical Power and Energy Systems 57 (2014) 358–365

resulting a lightning shield failure in the transmission system supported by tall towers. The second factor was observed in the timedomain analysis, the surge voltage levels on the tall line sections are significantly higher than a conventional line section with 40 m height. These two critical factor can result in failure in the system insulation by flashover and backflashover and consequently constant interruptions in the power transmission. An alternative lightning protection using metal-oxide surge arresters direct at the towers (line surge arresters) has been considered for several line sections with frequent incidence of atmospheric discharges. This lighting protection apparatus has been efficient for most of conventional transmission systems. Thus, the same lightning protection was proposed to deal with the shield failure of the new transmission system using tall towers. However, the transient simulations for both nominal 440-kV systems show that the proposed protection apparatus is efficient for protection of a conventional 440-kV line section. On the other hand, the same protection apparatus cannot maintain the surge voltage below the maximum operation voltage of 462 kV. This information represents an important conclusion on the lightning protection of the new transmission line. The current research presented several important conclusion on the frequency-dependent parameters of the line, electromagnetic transient behavior, possible failures in the project (lightning shield) and conclusion of an alternative lightning protection and limitations. The conclusions and discussions on the line protection were restricted to the lightning protection. However, complementary studies are necessary to provide a complete overview of this new technology. For example, a complete analysis of the relay protection is necessary to complement the proposed lightning protection. Another important and complex analysis is the comparison of results obtained from simulations with practical results obtained by measurements. These analyses are necessary to complement the results obtained in this research and to provide a deeper understanding of this emergent technology for power transmission. Acknowledgements Editor Tharam Dillon and two anonymous Reviewers which provide important suggestions and comments on this research.

365

This research was supported by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, Proc. 4570-11-1) and by Fundação de Amparo à Pesquisa do Estado de São Paulo (Proc. 13/ 00974-5). References [1] Technique catalog. Hebei Eissun Communication Equipment CO. [11.10]. [2] Boile GND, Nascimento RL. ‘‘Linhão Tucuruí – 1800 km de integração regional’’. Revista T & C Amazônia; 2010, vol. 8, no. 18, p. 58–62 [Portuguese]. [3] Frota WM. Melhorias estruturais de suprimento para os sistemas elétricos isolados de Manaus e Macapá. Revista T & C Amazônia; 2005, vol. 3, no. 6, p. 23–9 [Portuguese]. [4] Dommel HW. EMTP theory book. Canada: Vancouver; 1986. [5] Wedepohl LM, Wilcox DJ. Transient analysis of underground power transmission systems: system-model and wave propagation characteristics. IEEE Proc Gener Transm Distrib 1973;20(2):253–60. [6] Mingli W, Yu F. Numerical calculations of internal impedance of solid and tubular cylindrical conductors under large parameters. IEEE Proc Gener Transm Distrib 2004;151(1):67–72. [7] Carson JR. Wave propagation in overhead wires with ground return. Bell Syst Tech J 1926;5:539–56. [8] Rachidi F, Tkachenko SV. Electromagnetic field interaction with transmission lines: from classical theory to HF radiation effects. WIT Press; 2008. [9] Tesche FM, Ianoz MV, Karlsson T. EMC analysis methods and computational models. John Wiley and Sons Inc.; 1997. [10] Notaros BM. Eletromagnetismo, Person; 2011 [Portuguese]. [11] Kurokawa S, Pissolato J, Tavares MC, Portela CM, Prado AJ. Behavior of overhead transmission line parameters on the presence of ground wires. IEEE Trans Power Deliv 2005;20(2):1669–76. [12] IEC 60060-1. High-voltage test techniques – Part 1: general definitions and test requirements, 2nd ed.; 2009–2010. [13] Nguyen HV, Dommel HW, Marti JR. Direct phase-domain modelling of frequency-dependent overhead transmission lines. IEEE Trans Power Deliv 1997;12(3):1335–42. [14] Cigré Working Group 01 (Lightning) of Study Committee 33 (Overvoltages and insulation coordination): guide to procedures for estimating the lightning performance of transmission lines; 1991. [15] Chisholm WA, Chow YL, Srivastava KD. Lightning surge response of transmission towers. IEEE Trans Power Appl Syst 1983;102(9):3232–42. [16] Gutiérrez JA, Moreno P, Naredo JL, Bermúdez JL, Nucci CA, Rachidi F. Nonuniform transmission tower model for lightning transient studies. IEEE Trans Power Deliv 2004;19(2):490–6. [17] Grcev L, Rachidi F. On tower impedances for transient analysis. IEEE Trans Power Deliv 2004;19(3):1238–44. [18] IEC 60099-4. Surge arresters – Part 4: metal-oxide surge arresters without gaps for a.c. systems, 2nd ed.; 2004–2005. [19] Christodoulou CA, Ekonomou L, Fotis GP, Karampelas P, Stathopulos IA. ‘‘Parameters’ optimisation for surge arrester circuit models’’. IET Sci, Meas Technol 2010;4(2):86–92.